X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/5b407667ca47cf4d9752821fd49e5611737146d2..d887dee631335a8c44ef75250524e69e67fbc9b4:/librarian/dcparser.py diff --git a/librarian/dcparser.py b/librarian/dcparser.py index d99aaf0..78f231c 100644 --- a/librarian/dcparser.py +++ b/librarian/dcparser.py @@ -79,47 +79,67 @@ def as_unicode(text): else: return text.decode('utf-8') +def as_wluri_strict(text): + return WLURI.strict(text) + class Field(object): - def __init__(self, uri, attr_name, type=as_unicode, multiple=False, salias=None, **kwargs): + def __init__(self, uri, attr_name, validator=as_unicode, strict=None, multiple=False, salias=None, **kwargs): self.uri = uri self.name = attr_name - self.validator = type + self.validator = validator + self.strict = strict self.multiple = multiple self.salias = salias self.required = kwargs.get('required', True) and not kwargs.has_key('default') self.default = kwargs.get('default', [] if multiple else [None]) - def validate_value(self, val): + def validate_value(self, val, strict=False): + val = [v.strip() if v is not None else v for v in val] + if strict and self.strict is not None: + validator = self.strict + else: + validator = self.validator try: if self.multiple: - if self.validator is None: + if validator is None: return val - return [ self.validator(v) if v is not None else v for v in val ] + return [ validator(v) if v is not None else v for v in val ] elif len(val) > 1: raise ValidationError("Multiple values not allowed for field '%s'" % self.uri) elif len(val) == 0: raise ValidationError("Field %s has no value to assign. Check your defaults." % self.uri) else: - if self.validator is None or val[0] is None: + if validator is None or val[0] is None: return val[0] - return self.validator(val[0]) + return validator(val[0]) except ValueError, e: raise ValidationError("Field '%s' - invald value: %s" % (self.uri, e.message)) - def validate(self, fdict): + def validate(self, fdict, fallbacks=None, strict=False): + if fallbacks is None: + fallbacks = {} if not fdict.has_key(self.uri): if not self.required: - f = self.default + # Accept single value for single fields and saliases. + if self.name in fallbacks: + if self.multiple: + f = fallbacks[self.name] + else: + f = [fallbacks[self.name]] + elif self.salias and self.salias in fallbacks: + f = [fallbacks[self.salias]] + else: + f = self.default else: raise ValidationError("Required field %s not found" % self.uri) else: f = fdict[self.uri] - return self.validate_value(f) + return self.validate_value(f, strict=strict) def __eq__(self, other): - if isinstance(other, Field) and other.name == self.name: + if isinstance(other, Field) and other.name == self.name and other.uri == self.uri: return True return False @@ -144,9 +164,11 @@ class WorkInfo(object): __metaclass__ = DCInfo FIELDS = ( - Field( DCNS('creator'), 'authors', as_person, salias='author', multiple=True), + Field( DCNS('creator.expert'), 'authors_expert', as_person, salias='author', required=False, multiple=True), + Field( DCNS('creator.scenario'), 'authors_scenario', as_person, salias='author', required=False, multiple=True), + Field( DCNS('creator.textbook'), 'authors_textbook', as_person, salias='author', required=False, multiple=True), Field( DCNS('title'), 'title'), - Field( DCNS('type'), 'type', required=False, multiple=True), + Field( DCNS('type'), 'type', required=False), Field( DCNS('contributor.editor'), 'editors', \ as_person, salias='editor', multiple=True, default=[]), @@ -157,23 +179,26 @@ class WorkInfo(object): Field( DCNS('date.pd'), 'released_to_public_domain_at', as_date, required=False), Field( DCNS('publisher'), 'publisher'), + Field( DCNS('subject.competence'), 'competences', multiple=True, required=False), + Field( DCNS('subject.curriculum'), 'curriculum', multiple=True, required=False), + Field( DCNS('language'), 'language'), Field( DCNS('description'), 'description', required=False), Field( DCNS('source'), 'source_name', required=False), Field( DCNS('source.URL'), 'source_url', required=False), - Field( DCNS('identifier.url'), 'url', WLURI), + Field( DCNS('identifier.url'), 'url', WLURI, strict=as_wluri_strict), Field( DCNS('rights.license'), 'license', required=False), Field( DCNS('rights'), 'license_description'), ) @classmethod - def from_string(cls, xml): + def from_string(cls, xml, *args, **kwargs): from StringIO import StringIO - return cls.from_file(StringIO(xml)) + return cls.from_file(StringIO(xml), *args, **kwargs) @classmethod - def from_file(cls, xmlfile): + def from_file(cls, xmlfile, *args, **kwargs): desc_tag = None try: iter = etree.iterparse(xmlfile, ['start', 'end']) @@ -194,14 +219,14 @@ class WorkInfo(object): # if there is no end, Expat should yell at us with an ExpatError # extract data from the element and make the info - return cls.from_element(desc_tag) + return cls.from_element(desc_tag, *args, **kwargs) except XMLSyntaxError, e: raise ParseError(e) except ExpatError, e: raise ParseError(e) @classmethod - def from_element(cls, rdf_tag): + def from_element(cls, rdf_tag, *args, **kwargs): # the tree is already parsed, so we don't need to worry about Expat errors field_dict = {} desc = rdf_tag.find(".//" + RDFNS('Description')) @@ -214,9 +239,9 @@ class WorkInfo(object): fv.append(e.text) field_dict[e.tag] = fv - return cls(desc.attrib, field_dict) + return cls(desc.attrib, field_dict, *args, **kwargs) - def __init__(self, rdf_attrs, dc_fields): + def __init__(self, rdf_attrs, dc_fields, fallbacks=None, strict=False): """rdf_attrs should be a dictionary-like object with any attributes of the RDF:Description. dc_fields - dictionary mapping DC fields (with namespace) to list of text values for the given field. """ @@ -225,16 +250,14 @@ class WorkInfo(object): self.fmap = {} for field in self.FIELDS: - value = field.validate( dc_fields ) + value = field.validate(dc_fields, fallbacks=fallbacks, + strict=strict) + if field.multiple: + value = getattr(self, 'prop_' + field.name, []) + value setattr(self, 'prop_' + field.name, value) self.fmap[field.name] = field if field.salias: self.fmap[field.salias] = field - self.validate() - - def validate(self): - self.url.validate_language(self.language) - def __getattribute__(self, name): try: field = object.__getattribute__(self, 'fmap')[name] @@ -245,7 +268,7 @@ class WorkInfo(object): if not field.multiple: raise "OUCH!! for field %s" % name - return value[0] + return value[0] if value else None except (KeyError, AttributeError): return object.__getattribute__(self, name) @@ -344,17 +367,25 @@ class BookInfo(WorkInfo): Field( DCNS('audience'), 'audiences', salias='audience', multiple=True, required=False), - Field( DCNS('subject.period'), 'epochs', salias='epoch', multiple=True), - Field( DCNS('subject.type'), 'kinds', salias='kind', multiple=True), - Field( DCNS('subject.genre'), 'genres', salias='genre', multiple=True), - + Field( DCNS('subject.period'), 'epochs', salias='epoch', multiple=True, + required=False), + Field( DCNS('subject.type'), 'kinds', salias='kind', multiple=True, + required=False), + Field( DCNS('subject.genre'), 'genres', salias='genre', multiple=True, + required=False), + Field( DCNS('contributor.translator'), 'translators', \ as_person, salias='translator', multiple=True, default=[]), - Field( DCNS('relation.hasPart'), 'parts', WLURI, multiple=True, required=False), - - Field( DCNS('relation.cover_image.url'), 'cover_url', required=False), - Field( DCNS('relation.cover_image.attribution'), 'cover_by', required=False), - Field( DCNS('relation.cover_image.source'), 'cover_source', required=False), + Field( DCNS('relation.hasPart'), 'parts', + WLURI, strict=as_wluri_strict, multiple=True, required=False), + Field( DCNS('relation.isVariantOf'), 'variant_of', + WLURI, strict=as_wluri_strict, required=False), + Field( DCNS('relation'), 'relations', + WLURI, strict=as_wluri_strict, multiple=True, required=False), + + Field( DCNS('relation.coverImage.url'), 'cover_url', required=False), + Field( DCNS('relation.coverImage.attribution'), 'cover_by', required=False), + Field( DCNS('relation.coverImage.source'), 'cover_source', required=False), )